Why We Became a Performance Design Consultancy

Design got cheap. Not bad. Cheap. Getting well-considered interfaces and a functional design system stopped being a differentiator sometime in the last two years. Any competent designer with the right stack can produce work that looks like the work, fast, and at a fraction of what it cost before.

That's not a complaint. It's a condition. And the condition is: if you're running a design studio, every conversation starts drifting toward price. Price conversations mean the value proposition isn't clear. Usually that means it needs rethinking. Performance design is where that rethinking landed.

We ran form-three as a general design consultancy for two years before the economics stopped making sense. What we do now looks different: the metric comes before the mock-up, and the work is harder to explain at a dinner party.

form-three started in 2023. Singapore, complex digital platforms, B2B and B2C. The founding idea was accurate at the time: Singapore had a shortage of senior designers who could get things done without hand-holding. Multidisciplinary, comfortable at any level of an organisation, useful in a room with a CFO or a founding team. Not a studio that sends work through an account manager. An actual design partner.

Venture building was strong in Singapore in 2023. Lots of startups, lots of platforms being built, lots of companies that needed exactly that. For about two years, the model worked.

What changed

Three things happened roughly in parallel, and together they were harder to absorb than any one would have been alone.

AI went from something we experimented with to roughly half the daily work. Prototyping, analysis, information architecture, image generation, code scaffolding, competitive audits. Things that used to take a week took a day. Things that justified certain billing rates became prompts. The economics shifted fast. Foundation Capital and Designer Fund's 2025 State of AI in Design report found that 84% of designers now use AI in the exploration phase. That's the phase where the billing used to sit.

Venture building in Singapore slowed. Total startup funding in Singapore dropped 52% across the first nine months of 2024, the lowest level since 2020. The companies that would have needed a long-term embedded design partner stopped forming at the same rate. An anchor client kept form-three moving through most of that stretch, but the founding model was getting harder to defend regardless.

And then the uncomfortable one: screens became a commodity. Getting competent design at scale is now fast and accessible to anyone with a budget and a brief. Which means the argument "we're worth more because we're better" runs into a harder question: better at what, exactly, and does that difference produce anything the client can measure?

What stayed valuable

I keep coming back to one client when I try to explain this. A publicly listed financial comparison group where we worked embedded across multiple platforms for a long stretch. Every design decision had a conversion metric attached to it. Apply rate, purchase rate, time to buy. Nothing went unmeasured. And the team kept getting smaller as AI absorbed the repeatable work: copy variants, image production, early-stage prototyping, analysis. Not because the design function was failing. Because the scope of what a human designer needed to do was contracting.

What actually moved numbers was none of the stuff that got automated. It was reading the data and knowing which question to ask next. Knowing that the drop-off on step four wasn't a UI problem but a trust problem, because the user hadn't seen a price yet and was being asked for personal details. That kind of read comes from sitting inside a business for months, not from receiving a brief on Monday and delivering screens on Friday.

Most designers don't sit there. They receive the brief, deliver the output, move on. The brief-to-deliverable model will keep existing, there will always be work like that, but the rate card for it is dropping fast. I've watched three senior freelancers I respect lose retainers in the last year. Good people, good work. The work just isn't scarce anymore.

Performance Design

We needed a name for what we were actually doing. "UX" didn't cover it. "Growth" didn't either. The term isn't new outside of digital. In architecture, performance-based design has been a discipline for decades. A building doesn't just need to look right. It needs to meet measurable criteria: energy load, structural tolerance, acoustic performance, thermal comfort. The architect holds all of those in tension with the brief, the site, and the budget. A beautiful building that fails its energy target is a failed building. Nobody in that field treats performance as an afterthought or something a separate team handles post-handover.

Digital product design hasn't caught up to that. The closest the industry has is GrowthUX and CRO (conversion rate optimisation), which get the measurement part right but scope it too narrowly. Both live inside growth teams. Both optimise isolated page elements: this button, this form, this onboarding step. Run the A/B test, pick the winner, move on. The success metric is typically click-through rate or statistical uplift on a single variant. The problem is that optimising one touchpoint often degrades another. A shorter form converts better but produces lower-quality leads. A more aggressive upsell increases ARPU but tanks retention two months later. Neither GrowthUX nor CRO has a framework for holding those trade-offs because neither is designed to look at the end-to-end decision system.

Performance Design does. Where traditional UX focuses on screens and components, and CRO focuses on isolated page elements, Performance Design works on the end-to-end decision system: how every step in a flow earns the right to the next one. It's a way of working where the balance between business objectives and user needs is explicit from the start, not assumed to be aligned. The job isn't to serve user needs and hope business results follow. It's to hold both in the frame and make the call, including when the business objective needs to carry more weight than a strictly user-centred approach would typically allow.

Most designers are trained to push back on business pressure. That instinct is often right. But "the business wants this and it conflicts with users" is a much more specific statement than it's usually treated as. Sometimes the business wants something, users adapt, and the metric goes up. Designers don't love hearing that. I didn't, for a long time.

Getting the balance right is most of the job. Not a fixed ratio. It shifts per project, per screen, sometimes per form field. Performance Design requires naming which side is under-weighted on any given decision and adjusting for it. Usually it's the business side. Sometimes you over-correct the other way and find out a week later from the data. You won't get it right every time. But if you're treating the balance itself as the design problem rather than something you settled in a kickoff meeting, you're closer than most teams get.

The work that follows is concrete: start with a friction audit to find what the funnel is losing and why, design against that specific problem, measure whether it moved. We spend more time in Looker than in Figma most weeks. The whole flow, not one touchpoint at a time.

flow-three

flow-three is what happened when we stopped saying yes to everything and started saying yes only to this.

We stopped taking general platform design work, visual refreshes, systems projects that didn't connect to a measurable outcome. What's left is embedded, data-oriented performance design: working on the gap between what a product currently does and what the business needs it to do.

I brought in Alexa and Chin-Han. Alexa heads up delivery. She worked with form-three before flow-three existed, and before that we overlapped at OnlinePajak. By the time we built this, we already knew how the other one works under pressure. More recently she was lead product designer at PropertyGuru, where a single upsell redesign drove over SGD $860K in additional annual revenue. That's not a case study talking point. It's the kind of outcome that makes the performance design argument easy to explain to a client who's sceptical.

Chin-Han leads growth. His background is the business side: co-founder of UNCANNY, growth director across APAC at Media.Monks. He brings the commercial lens that Alexa and I don't have natively. How to find clients, convert interest into work, build the growth mechanics that most designers leave to someone else, usually to their own cost.

Alexa and I can do the work but we're not great at finding it. Chin-Han can find it but he can't do it. That tension is the whole point. We argue about pricing, about scope, about which clients to chase. The arguments are usually where the good decisions come from.

The friction audits we run for clients follow a specific process: map the flow, score every touchpoint for friction across five dimensions, surface where the revenue impact actually sits, then design against those findings. We did that manually for long enough that the process became predictable. So we built it into a platform.

flow-three is now also a product. It scores each touchpoint for friction, maps the flow end to end, and translates the findings into estimated revenue at risk. The methodology, made into software. Self-serve for teams that want to run the audit themselves, consultancy for the ones who want us to do the work and own the outcome. Whether those two sides stay in balance or one pulls harder, I don't know yet. We'll find out. Private beta is opening soon. If that's interesting, get on the list.

On the name

People ask.

flow is the business objective. Better flow through a product means more users completing what the business needs them to complete. Not an aesthetic goal. A revenue one.

three is the system. Product, funnel, and journey. Where a user drops off is rarely just one of them. Where they convert is almost always the result of all three working together. Performance design that only touches one usually optimises it at the cost of another.

Lowercase because it looks better.

---

If you're measuring traffic but not friction, you're looking at the wrong number. We do a 30-minute friction read on one of your flows, no pitch attached. Book a slot.

Related: What performance design actually means · The OnlinePajak activation project: where the methodology got tested